This study evaluates the effect of hydraulic loading rate (HLR), aspect ratio, granular medium size and water depth on the removal of selected contaminants during the start up of horizontal subsurface flow reed beds (HFRBs). Experiments were carried out in a pilot-scale HFRB system comprising four pairs of lined beds of almost equal surface area (54-56 m(2) each bed), with the following aspect ratios: 1:1, 1.5:1, 2:1 and 2.5:1. The size of the granular medium of each pair varied from coarse granitic gravel ( D(60) = 10mm, C(u) = 1.6) to small granitic gravel (D(60) = 3.5 mm, C(u) = 1.7). The beds of the pair with longest aspect ratio were made shallower (0.27 m) than the rest (0.5m) The system was sampled weekly from May 2001 to January 2002. The results indicate that HLR and water depth are determining factors in the performance of the HFRBs. Beds with a water depth of 0.27 m removed more COD (70-80%), BOD(5) (70-85%), ammonia (40-50%) and dissolved reactive phosphorus (DRP) (10-22%) than beds with a depth of 0.5m (60-65% for COD, 50-60% for BOD(5), 25-30% for ammonia, and 2-10% for DRP). The higher efficiency observed shallower beds was related to their less reducing conditions (average redox potential (E) ranging from -351 to -338 mV) than beds with a depth of 0.5m (-390 to -358 mV). The difference in E status between two bed types seems to lead to differences in the biochemical processes. In fact, denitrification was estimated to be a significant reaction in shallower beds.